KR980013432A - Coding method and apparatus for object image - Google Patents

Abstract

The present invention improves the coding efficiency by moving a macroblock partitioning a VOP in MPEG-4 to an information amount reduction position according to shape information of an object image.

The present invention defines a VOP formed according to an image of an object, and finds a position for reducing the amount of information while moving the start point of the X axis and the Y axis of the macro block of the VOP defined above. The information amount reduction position is determined as an optimal X-axis and Y-axis grid start point of the X-axis and Y-axis grid points where the number of macroblocks in which an image of the object exists is minimized and the number of macro blocks is minimized. After reforming the macroblock according to the starting points of the X-axis and Y-axis grids, the shape coding unit encodes the shape information. Then, the X-axis and Y-axis grid start points of the macroblocks shape-coded by the shape encoding unit are moved, and the positions where the number of macroblocks in which the image of the object exists are minimized are determined as the optimal X-axis and Y- The macroblock is re-formed according to the determined optimal X-axis and Y-axis grid starting points to perform motion estimation, motion compensation, and intra-object coding.

Description

Method and apparatus for encoding object image and method and apparatus for encoding object image

The present invention relates to a method and apparatus for moving a macroblock for dividing a VOP (Video Object Plane) in MPEG (Noving Picture Experts Group) -4 to an information amount reduction position according to the position of shape information for an object image such as an object and an area, And the object image to be enhanced is encoded by an encoding method and an encoding device.

MPEG-4, which is currently being standardized, is based on the concept of VOP (Video Object Plane).

Here, the VOP has a basic skeleton in which, when an object image such as a predetermined object and an area exists on an image screen, the image of the object is separated and the images of the separated object are separately encoded.

Such a VOP has the advantage of being able to freely synthesize or disassemble a natural image or an artificial image as a unit of an object image, and is becoming a basis for processing an image of an object in the fields of computer graphics and multimedia.

FIG. 1 is a block diagram showing the configuration of a VM (Verification Model) encoder that is primarily determined by the International Organization for Standardization (ISO / IEC JTC1 / SC29 / WG11 MPEG96 / N1172 JanuarY).

Here, the VOP formation unit 11 forms a VOP when a video sequence to be transmitted or stored is input.

The formation of the VOP is performed by dividing a background screen into a plurality of object images on one screen and a different VOP for each object image into a background image and respective object images, The smallest rectangle containing the object image is defined as VOP.

FIG. 2 shows an example in which one V0P is formed by setting an image of "cat" as an image of an object.

Here, the horizontal size of the VOP is defined as the VOP width, and the vertical size is defined as the VOP height.

The formed VOP is partitioned into M × N macroblocks having M pixels and N pixels on the X axis and Y axis, respectively, with the upper left corner as the grid starting point. For example, 16x16 macroblocks each having 16 pixels on the X axis and the Y axis.

In this case, when the X and Y axis pixels of the macroblocks formed on the right and bottom sides of the VOP are not M and Y, respectively, the magnitude of the VOP is expanded so that the X and Y axis pixels of each macroblock are both M And N pieces.

The M and N are set to an even number in order to perform encoding in units of sub-blocks in an object coder to be described later.

Each VOP formed in the VOP forming unit 11 is input to each of the VOP encoding units 12A, 12B, 12C, ... and is encoded for each V0P, multiplexed by the multiplexer 13, and transmitted as a bit stream.

FIG. 3 is a block diagram showing a configuration of a VM decoder which is primarily determined by an organization under international standards.

The encoded signal of the VOP, which is encoded through the encoder and transmitted as the bit stream, is demultiplexed by VOP in the demultiplexer 21 of the VM decoder.

The output signals of the VOP decoders 22A, 22B, 22C, ... are synthesized in a synthesizing unit 23, and the output signals of the VOP decoders 22A, 22B, 22C, The image is displayed on the screen.

FIG. 4 is a block diagram showing the configuration of VOP encoding units 12A, 12B, 12C,... Of VM encoders which are primarily determined by the international standards organization.

A VOP for each object image formed in the VOP forming unit 11 is input to a motion estimation unit 31 to estimate motion in units of macroblocks.

The motion information estimated by the motion estimation unit 31 is input to a motion compensation unit 32 to compensate for motion.

The motion compensated VOP in the motion compensation unit 32 is input to the adder 33 together with V0P formed in the V0P forming unit 11 to detect a difference value and the difference value detected by the adder 33 is Object internal encoding unit 34 and the internal information of the object is encoded in units of sub-blocks of the macro-block.

For example, the intra-object information is encoded after the X and Y axes of the macroblock are subdivided into 8 × 8 sub-blocks each having 8 pixels with M / 2 × N / 2.

The motion compensated V0P in the motion compensation unit 32 and the internal information of the object coded by the object inner coding unit 34 are inputted to the adder 35 and added thereto and the output signal of the adder 35 is supplied to the previous V0P V0P of the previous screen is input to the detection unit (Previous Reconstructed VOP 36). V0P of the previous screen detected by the previous V0P detection unit 36 is input to the motion estimation unit 31 and the motion compensation unit 32 And is used for motion estimation and motion compensation.

The VOP formed in the VOP forming unit 11 is input to a shape coding unit 37, and shape information is encoded.

Here, the output signal of the shape coding unit 37 varies depending on the field to which the VOP coding unit 12, 12A, 12B, ..., is applied, and the output signal of the shape coding unit 37 is represented by a dotted line As shown, to the motion estimation unit 31, the motion compensation unit 32, and the object intra-coding unit 34 to perform motion estimation, motion compensation, and coding of internal information of the object.

The motion information estimated by the motion estimating unit 31, the internal information of the object encoded by the object inner encoding unit 34 and the shape information encoded by the shape encoding unit 37 are multiplexed by the multiplexer 38, 39 to the multiplexer 13 of the second degree and transmitted as a bit stream.

In this MPEG-4, when the VOP is formed, M × N macroblocks having M pixels and N pixels are respectively formed on the X axis and the Y axis from the VOP grid start point of the upper left corner.

At this time, if the image of the object, that is, the pixel having the shape information of the object exists in a large number of macroblocks, the coding efficiency becomes low because the number of macroblocks to be encoded with shape information increases.

Therefore, according to the conventional technique in which the image of the object is formed in units of M × N macroblocks by setting the upper left corner as the grid start point, the number of macroblocks to be encoded in the shape information of the object is increased, .

For example, in the drawing of FIG. 2 in which "cat" is displayed as an object, the number of macroblocks in which pixels of "cat" are located is 20, and 7 to 8 macroblocks include only pixels of small size. Technology has to encode all 20 macroblocks in which the "cat " pixel is located, resulting in low coding efficiency.

Since the amount of information output from the shape encoding unit 37 is large even when the shape information encoding unit 37 performs motion estimation, motion compensation, and intra-object encoding of the shape information encoded signal, motion estimation, motion compensation, The amount of information is large.

Accordingly, an object of the present invention is to provide a method and apparatus for coding an object image, which moves a macroblock of a VOP to an information amount reduction position according to shape information of an object image, and then performs shape information coding to improve shape information coding efficiency.

It is another object of the present invention to improve the coding efficiency and reduce the amount of information by performing motion estimation, motion compensation, and object inner coding after moving a macroblock of a VOP coded signal in which shape information is encoded in a shape coding unit to an information amount reduction position And to provide a coding method and a coding apparatus for an object image.

In order to achieve the above object, the present invention defines a VOP formed according to an image of an object, and finds a position where the amount of information is reduced while moving the start point of the X axis and the Y axis of the macro block of the VOP defined above.

For example, the X-axis and Y-axis grid starting points, at which the number of macroblocks in which the image of the object is present is minimized, are determined as the optimal X-axis and Y-axis grid starting points After the macro block is reformed according to the determined optimal X-axis and Y-axis grid start points, shape information is encoded in the shape coding unit.

The position where the number of macroblocks in which the image of the object exists is minimized while moving the X axis and Y axis grid start points of the macroblocks shape-coded by the rig-shape coding unit as the optimal X axis and Y axis grid start points The macroblock is re-formed according to the determined optimal X-axis and Y-axis grid starting points to perform motion estimation, motion compensation, and intra-object coding.

Hereinafter, a method and a device for encoding an object image according to the present invention will be described in detail with reference to the accompanying drawings of FIGS. 5 to 18, wherein the same parts as those of the conventional art are denoted by the same reference numerals.

FIG. 5 is a block diagram showing an embodiment in which a shape adaptive block dividing unit of the encoder of the present invention is configured in a VOP coding unit 12A, 12B, 12C,... Of a VM encoder which is primarily determined by an international standard organization .

As shown in the figure, the present invention includes a first shape adaptive block dividing unit 40 between the V0P forming unit 11 and the shape coding unit 37 to move the X and Y axis grid start points of the macroblocks And determines an information amount reduction position according to the position of the object image of the VOP.

The optimal X-axis and Y-axis grid start points are determined as the optimum X-axis and Y-axis grid start points, and the shape encoding unit 37 ).

The shape coding unit 37 reforms the macroblock according to the optimal X-axis and Y-axis grid starting points output from the first shape adaptive block dividing unit 40, and encodes the shape information of the object.

And a second shape adaptive block dividing unit 41 between the shape coding unit 37 and the motion estimation unit 31, the motion compensation unit 32 and the object inner coding unit 34.

The second shape adaptive block dividing unit 41 finds and outputs the information amount reduction position while moving the X axis and Y axis grid start points of the macroblocks in which the shape information is encoded by the shape coding unit 37. [

That is, the second shape adaptive block dividing unit 41 determines the starting points of the X-axis and Y-axis grids as the optimal X-axis and Y-axis grid starting points, where the number of macro blocks in which the image of the object exists is minimized.

The motion estimation unit 31, the motion compensation unit 32, and the object inner coding unit 34 determine the macroblocks according to the optimal X-axis and Y-axis grid starting points determined by the second shape adaptive block dividing unit 41 And motion estimation, motion compensation, and intra-object information are encoded.

FIG. 6 is a circuit diagram showing an embodiment of the first and second shape adaptive block dividing sections 40 and 41 of FIG. 5. Reference numeral 51 denotes an address generation controller for moving the address start position at which the address is to be generated at regular intervals within the size of the X and Y axes of one macro block.

Reference numeral 52 denotes an address generator for generating an X-axis and a Y-axis address so that an image of an object is divided into macroblocks in accordance with an address start position outputted by the address generation controller 51 and output sequentially. Reference numeral 53 denotes a memory for sequentially storing object images having shape information and / or shape information to be inputted and dividing them into macroblocks according to the addresses generated by the address generator 52 and outputting them.

The address generator 52 determines the X-axis size of the macro block based on the size information of the input object image as shown in FIG. 7, and the Y-axis size determination unit 522 ) Determines the Y-axis size of the macroblock.

Here, when the sizes of the X and Y axes of the macroblock are the same, the size of the macroblock can be determined by using only one size determining unit.

The macroblock address generator 523 in the address generator 52 determines the X axis size and the Y axis size of the macroblock determined by the X axis size determiner 521 and the Y axis size determiner 522, The X axis and the Y axis address of the macroblock of the object image stored in the memory 53 are discriminated on the basis of the address start position outputted from the address generation controller 51, Y-axis addresses are sequentially generated.

The memory 53 divides the stored image into macroblocks in order according to the X axis and Y axis addresses sequentially generated by the macroblock address generator 523. For example, as shown in FIG. 8, object images of one macroblock are sequentially output, and object images of the next macroblock are sequentially output.

Reference numeral 54 denotes a block number counter for counting the number of macroblocks in which the image of the object exists among the output signals of the memory 53. [

The block number counter 54 counts the clock signal by the macro block counter 541 as shown in FIG. 9, and distinguishes the macro blocks. The shape information presence determiner 542 of the block number counter 54 divides the macroblock of the object image output from the memory 53 according to the output signal of the macroblock counter 541 to determine whether the object image exists And the adder 543 adds the determination signal of the shape information existence determination unit 542 to output the number of macroblocks in which the image of the object exists.

Reference numeral 55 denotes a minimum macroblock grid selector for selecting the X axis and Y axis grid start points, which are obtained by counting the smallest number of macroblocks out of the number of macroblocks counted by the block number counter 54, as information amount reduction positions.

The minimum macroblock grid selector 55 stores the count value when the count of the block number counter 54 is completed and controls the address generation controller 51 to set the starting position of the X and Y axis addresses Is generated within a size of the X and Y axes of the macroblock at regular intervals. That is, the minimum macro-block grid selector 55 moves the X-axis and Y-axis address start positions, and then, when the operation of counting the number of macroblocks in which an image of the object exists is completed, the address generation controller 51 And the starting position of the X-axis and Y-axis addresses is moved.

When the operation of counting the number of macroblocks in which an image of an object exists is completed while moving the starting position of the X and Y axis addresses within the size of the X and Y axes of one macroblock, The selector 55 determines the start positions of the X-axis and Y-axis addresses in which the smallest number of the macroblocks counted from the counted number of the macroblocks is counted, and sets the determined X-axis and Y-axis address start positions to the optimum X axis and Y axis grid start positions and outputs them.

FIG. 10 is a signal flow chart showing the encoding method of the present invention.

First, in step S11, a VOP formed in the VOP forming unit 11 is defined. For example, a VOP is defined for the image of "cat" shown in FIG. 11, and the VOP width and VOP height are determined.

In the next step S12, macroblocks having M × N pixels are partitioned as shown by the dotted line in FIG. 11 with reference to the grid start point at the upper left of the defined V0P.

If the macroblock is partitioned, the information amount reduction position is determined while moving the grid start point of the macroblock in step S13. For example, as shown by a solid line in FIG. 11, an optimum grid start point in which the object image exists in the smallest number of macro blocks is determined as the information amount reduction position.

If the information amount reduction position is determined in step S13, the shape information is encoded in the shape coding unit 37 according to the information amount reduction position in the next step S14.

In the next step S15, as in the step S13, the macroblock coding the shape information in the shape encoding unit 37 moves the grid start point and determines the information amount reduction position And outputs it.

In step S16, the motion estimation unit 31 estimates motion in macroblock units on the basis of the information amount reduction position output in step S15, motion compensation is performed in the motion compensation unit 32, The encoding unit (34) encodes the object inside position in sub-block units.

In the next step S17, the encoded shape information, motion information, and object inside information are output through the multiplexer 38 and the buffer 39. [

FIG. 12 is a signal flow diagram illustrating an embodiment of finding an information amount reduction position of a macroblock in steps S13 and S15 according to the encoding method of the present invention.

First, in step S21, in order to find an optimal X-axis and Y-axis grid start point, which is an information amount reduction position where the number of macroblocks having an image of an object having shape information and / or shape information is minimized, Initialize the grid start point (XM) (YN).

The initial values of the X and Y-axis grid starting points (XM) (YN) are given by XM = O, YN = 0. In the next step S22, the number of macroblocks in which an image of the object exists is counted based on the X axis and Y axis grid start points initialized in step S21.

Taking the macroblock of the VOP shown by the dotted line in the drawing of FIG. 11 as an example, the number of macroblocks in which the image of "cat" exists from the first row in the X- Cat "are present in all 20 macroblocks.

When the operation of counting the number of macroblocks is completed in the step S22, the number of the counted macroblocks is stored in the step S23.

In the next step S24, the X-axis grid start point XM is moved by a predetermined distance K from the first macro-block to the X-axis to reconstruct the macro block. In step S25, the X- It is judged whether or not it has been moved M times or more.

That is, it is determined whether or not the X-axis grid start point XM has moved to the X-axis size or more of one macro block.

If the X-axis grid start point XM has not been moved in the X-axis more than M times in step S25, the above-described steps S22-S25 are performed to set the X-axis grid start point XM at a predetermined interval K), and counts and stores the number of macroblocks in which an image of the object exists.

If the X-axis grid start point XM has been moved M times or more in the X-axis in step S25, the X-axis grid start point XM counting the smallest number of macroblocks in step S26 is set to the optimal X- It is determined as the grid start point (XOM).

In step S27, the macro block is reconstructed to the optimum X-axis grid start point (XOM) and the Y-axis grid start point (YN = 0) determined and the number of macro blocks in which the image of the object exists is counted.

When the operation of counting the number of macroblocks is completed in step S27, the number of counted macroblocks is stored in step S28.

In the next step S29, the Y-axis grid start point YN is shifted by a predetermined distance L along the Y-axis and whether or not the Y-side grid start point YN is shifted N times or more is determined in step S30 .

That is, it is determined whether the Y-axis grid start point YN has moved to a size larger than the Y-axis size of one macroblock.

If it is determined in step S30 that the Y-axis grid start point YN has not been moved N times or more at regular intervals L on the Y-axis, the above steps S27 to S30 are repeated to determine the optimal X- Axis grid start point YN is moved by a predetermined interval L based on X0M, and the number of macroblocks in which an image of the object exists is counted and stored.

If the Y-axis grid start point YN is shifted N times or more by a predetermined interval L in step S30, the Y-axis grid start point XM that counts the smallest number of macro blocks in step S31 is the optimal Y axis grid start point YON and outputs the optimal X axis grid start point XOM and Y axis grid start point YON determined in steps S26 and S31 to the information amount decrease position.

That is, in one embodiment of the present invention shown in FIG. 12, the entire grid is moved M times by a predetermined interval (K) on the X axis, and the X axis grid start point (XM) as the optimal X-axis grid start point (XOM). Axis grid start point (Y-axis) that counts the smallest number of macroblocks in which the image of the object exists, while moving the entire grid to the Y-axis N times at regular intervals (L) with reference to the determined optimal X- YN) as the optimal Y-axis grid start point (YON), and outputs the determined optimum X-axis grid start point (XOM) and optimal Y-axis grid start point (YON) to the information amount decrease position.

Therefore, an embodiment of the present invention moves the grid starting point M times on the X axis and N times on the Y axis, thereby moving the grid starting point XM, YN as a whole M + N times, Find and print the starting point (XOM, YON).

FIG. 13 is a signal flow diagram illustrating another embodiment for finding an information amount reduction position according to the encoding method of the present invention.

The X axis and Y axis grid start points XM (X) and Y (Y) are searched to find the optimal X and Y axis grid start points (XOM, YON) in which the number of macroblocks in which the image of the object on which V0P is formed is minimized, YN) is initialized to XM = O, YN = 0.

In step S42, the number of macroblocks in which the image of the object exists is counted based on the X-axis and Y-axis grid starting points (XN = 0) (YN = 0) initialized in step S41, S43), the number of the counted macroblocks is stored.

In the next step S44, the X-axis grid start point XM is shifted by a predetermined distance K on the X-axis. In step S45, the X-axis grid start point XM is moved on the X-axis M times or more, It is judged whether or not it has moved to a size larger than the size of one macroblock.

If the X-axis grid start point XM has not been moved M times or more in the X-axis in the step S45, the steps S42 to S45 are performed to set the X-axis grid start point XM at the predetermined interval K ), And repeats counting and storing the number of macroblocks in which an image of the object exists.

If the X-axis grid start point XM is moved more than M times in the X-axis in step S45, the Y-axis grid start point YN is moved in the Y-axis direction by a predetermined distance L in step S46.

In the next step S47, it is determined whether the Y-axis grid start point YN has been moved N times or more at a predetermined interval (L) along the Y-axis.

If the Y-axis grid start point YN is not moved in the Y-axis N times or more in the step S47, the steps S42 to S47 are performed to move the Y-axis grid in the Y-axis by a predetermined distance L The X-axis grid is moved along the X axis by a predetermined interval (K) within the size of one macroblock, and the number of macroblocks in which the image of the object exists is counted and stored.

If the Y-axis grid start point YN is moved N times or more in the Y-axis in step S47, the X-axis grid start point XM in which the smallest number of macroblocks in which the image of the object exists is counted in step S48. And Y-axis grid starting point (YN) are determined as optimum X-axis grid starting point (XOM) and Y-axis grid starting point (YON), and the determined optimum X- .

That is, in another embodiment of the present invention shown in FIG. 13, the entire grid starting point is shifted M times on the X axis by a predetermined interval (K), then moved by a predetermined distance L on the Y axis, The number of macroblocks in which an image of an object exists is repeated while repeating the operation of moving the object by M by a predetermined interval K on the X axis and moving the object by a predetermined distance L along the Y axis, And outputs the counted X-axis and Y-axis grid start points (XM) (YN) to the information amount reduction position.

Therefore, another embodiment of the present invention shown in FIG. 13 counts the smallest number of macroblocks while moving the X-axis and Y-axis grid starting points XM (YN) by a predetermined interval (K) (L) Outputs the X and Y axis grid start points (XM) (YN).

In the embodiments of FIGS. 12 and 13, the grid starting points are shifted by a predetermined interval (K) in the X-axis and then moved by one-point intervals (L) in the Y-axis to obtain optimum X- and Y- , YON) are described as an example.

However, in carrying out the present invention, as shown in the parentheses in the drawings of FIGS. 12 and 13, it is preferable to move the X-axis and the Y-axis at regular intervals (L) You can also find and print the Y axis grid start point (XOM, YON).

FIG. 14 is a signal flow diagram illustrating yet another embodiment for finding an information amount reduction position according to the encoding method of the present invention.

The X and Y axis grid start points XM and YN are all set to '0' in order to find the optimal X and Y axis grid start points XOM and YON in which the number of macro blocks in which the image of the object is located is minimized in step S51. O '.

In the next step S52, the number of macroblocks in which the image of the object exists is counted at the X-axis and Y-axis grid starting points XM and YN that have been initialized. In step S53, do.

In the next step S54, it is determined whether the X-axis grid start point XM is shifted M times on the X axis and the Y-axis grid start point YN is shifted N times on the Y side.

If it is determined in step S54 that the X-axis grid start point X14 has not been moved to the X-axis M times or more, or if the Y-axis grid start point YN has not been moved to the Y-axis N times or more, The Y-axis grid starting point XM (YN) is moved in a zigzag direction within a size of one macroblock by a predetermined interval (K) (L), and steps S52 to S55 are performed, The number of blocks is counted and stored, and the X-axis and Y-axis grid start points (XM) (YN) are repeatedly shifted in the zigzag direction.

Here, there are two methods of moving the X-axis and Y-axis grid starting points XM and YN in the zigzag direction by a predetermined distance K (L). For example, it can move along the direction of the arrow shown in FIG. 15 (A) or along the direction of the arrow shown in FIG. 15 (B).

If the X-axis grid start point XM is shifted M times on the X axis and the Y-axis grid start point YN is shifted N times on the Y axis in step S54, the smallest number of macro blocks is calculated in step S56 (XOM) and the Y-axis grid start point (YN) as the optimal X-axis grid start point (XOM) and Y-axis grid start point (YON) And Y-axis grid start point (YON) to the information amount reduction position.

That is, another embodiment of the present invention shown in FIG. 14 counts the number of macroblocks in which an image of an object exists while moving the grid starting point zigzag within the size of the X axis and the Y axis of one macro block.

Therefore, another embodiment of the present invention shown in FIG. 14 moves the X-axis and Y-axis grid starting points XM (YN) by a predetermined interval K (L) And outputs the counted X-axis and Y-axis grid start points (XM) (YN).

An example of reforming the macroblock according to the optimal grid starting point XM (YN) determined according to the embodiment of the above-mentioned FIG. 12 to FIG. 14 is shown by a solid line in FIG.

Here, the number of macroblocks in which an image of "cat" exists as a result of reconstructing a macroblock by moving the optimal X-axis and Y-axis grid starting points (XON) It can be seen that from the first row on the X axis, 2,3, 4, 3, and 5 are reduced to 17 in 20 macroblocks.

FIG. 16 is a signal flow diagram showing another embodiment for finding the position of a reduced amount of information of a macroblock according to the coding method of the present invention.

In step S61, the X-axis and Y-axis grid start points (XN, YN) are initialized to "0" to find the information amount reduction position where the number of macroblocks in which the image of the object exists is minimized.

In step S62, the number of macroblocks in which the image of the object exists in the X axis and Y axis grid starting points (XM = 0, YN = 0) initialized in step S61 are all counted, ) Stores the number of macroblocks counted in the step S62.

In the next step S64, the Y-axis grid starting point YN is shifted by a predetermined distance L on the Y-axis and whether or not the Y-axis grid starting point YN is shifted N times or more in the Y-axis in step S65 is determined .

If the Y-axis grid start point YN is not moved on the Y-axis N times or more, the Y-axis grid starting point YN is shifted by a predetermined distance L on the Y-axis by performing the steps S62 to S65, Is counted and stored.

If the Y-axis grid start point YN is moved N times or more in the Y-axis at a predetermined interval L in step S65, the Y-axis grid start point YN ) As an optimal Y-axis grid start point (YON).

If the optimal Y-axis grid start point YON counting the smallest number of macroblocks is determined in step S66, the grid is calculated based on the optimal Y-axis grid start point YON determined in step S67 And the macro block in the current X-axis row in which the image of the object exists is counted in step S68.

That is, as shown in (A) of FIG. 18, a macroblock in which an image of an object exists is counted from among the macroblocks of row X (1) which is the first row in the X axis.

In the next step S69, the number of the counted macroblocks is stored.

The grid start point XM of the X (1) row is shifted by a predetermined distance K on the X axis in the next step (S70) and the grid start point XM of the X (1) row is shifted to the X side It is determined whether it has moved M times or more by a predetermined interval (K).

If the X-axis grid start point XM of the X (1) row has not been moved to the X side M times or more in the short term (S71), the above steps (S68 to S71) (XM) is moved in the X axis by a predetermined interval (K), and macroblocks in which the image of the object exists in the X (1) row are counted and stored.

If the grid start point XM of the row X (1) is shifted more than M times by a predetermined interval (K) in the step S71, the smallest number of the macroblocks counted to the present in the step S72 The grid start point XM of the X (1) row counted from the start X (1) is determined as the optimum X (1) row grid start point X1M.

In the next step S73, it is determined whether it is the last row on the X axis. If it is not the last row on the X axis, the process moves to the next row on the X axis in step S74 and performs the above steps S58 to S74.

This operation is repeated to sequentially move the X-axis rows to X (1), X (2), X (3), X (4) and X (5) X (1), X (2), X (3), X (4) and X (5) X (3), X (4) and X (5) are determined as the starting points (X1M, x2M, ...) of the row grid.

If it is the last row in the X axis in the step S74, the optimal Y axis grid starting point YON and X (1), X (2), X (3), X (4) and (X1M, X2M, ...) of the X (5) rows to the information amount reduction position.

FIG. 17 is a signal flow diagram illustrating another embodiment for finding a position of a reduced amount of information of a macroblock according to the encoding method of the present invention.

In step S81, the X and Y axis grid start points XM and YN are initialized to '0' in order to find an information amount decrease position where the number of macroblocks in which an image of an object exists is minimized.

In the next step S82, a macroblock in which the image of the object exists is counted from among the macroblocks of the current X-axis row. That is, a macroblock in which an image of an object exists is detected and counted among the macroblocks in the row X (1).

When the counting of the macroblock is completed in the step S82, the number of the macroblocks counted in the step S83 is stored.

In the next step S84, the grid start point XM of the X (1) row is shifted by a predetermined distance K on the X axis and the grid start point XM of the X (1) K) by M times or more.

If the grid start point XM of the X (1) row has not been moved M times or more at regular intervals (K) in the step S85, the above steps (S82-S85) The grid start point XM is moved by a predetermined interval K, and the number of macroblocks in which an image of an object exists is counted and stored.

If the X-axis grid start point XM of the row X (1) is shifted by M times or more at regular intervals K in the X-axis in the step S85, the X (1) row counted up to the present in the step S86 The grid start point XM of the X (1) row in which the smallest number of macroblocks is counted among the macroblocks is determined as the optimum X (1) row grid start point XM.

In the next step S87, it is determined whether or not the last row is the X axis. If it is not the last row in the X axis, the next row of the X axis, namely X (2), X (3), X X (2), X (3), X (4), and X (2) obtained by counting the smallest number of macroblocks in which the image of the object exists are sequentially performed in the step S82- It is repeated to sequentially determine the grid start point XM of the X (5) row to the optimum X (2), X (3), X (4) and X (5) row grid start points (X1M, X2N, do.

If it is the last row in the X-axis in the step S87, it is determined in step S89 whether the Y-axis grid start point YN has been moved N times or more at a predetermined interval L in the Y-axis.

If the Y-axis grid start point YN has not moved N times in the Y-axis at a predetermined interval L in step S89, the Y-axis grid start point YN is shifted to the Y-axis at a predetermined interval (step S90) L), and repeats the steps S82 to S90.

That is, the Y-axis grid starting point YN is shifted by a predetermined distance L in the Y-axis and X (1), X (2), X (3), X The number of macroblocks in which an image of an object exists is sequentially counted while moving the X (5) row grid start point X (4) and the X (5) row grid point sequentially at predetermined intervals (K) (1), X (2), X (3), X (4), X (2), X (3), X And X (5) row grid start points (X1M, X2M, ...).

If the Y axis grid start point YN is shifted N times or more by a predetermined interval L in the step S90, then in the next step S91, the optimum X The total number of macroblocks counted at the row grid start point (X1N, X2N, ...) of X (1), X (2), X (3), X (4) and X (5)

In the next step S92, the Y-axis grid start point YN in which the smallest number of macroblocks is counted as a result of summing is determined, and the determined Y-axis grid start point YN is set as an optimal Y-axis grid start point YON . The X (1), X (2), X (3), and X (5) row grid start points X1M, X2M, ...) are determined as optimum X (1), X (2), X (3), X (4) and X (5) row grid start points (X1M, X2M, ...) Axis grid start point YON and X (1), X (2), X (3), X (4) and X (5) row grid start points X1M, X2N,.

The result of reconstructing the macroblock according to the optimum grid starting point found in the embodiments of FIGS. 16 and 17 is as shown in FIG. 18 (a).

In the embodiment shown in FIGS. 16 and 17, the Y-axis grid start point YN is moved and then the X-axis grid start point XM is moved to find the information amount decrease position.

In carrying out the present invention, the X-side grid start point XM and the Y-axis grid starting point YN are interchanged as shown in the parentheses in FIGS. 16 and 17, so that the information amount of the image of the object in the smallest macroblock You can also find the reduction position.

The result of reconstructing the macro block at the optimum grid starting point found by interchanging the X-axis grid start point XM and the Y-side grid start point YN is as shown in FIG. 18 (B).

In the embodiment of the present invention, the predetermined interval K (L) for moving the X-axis and Y-axis grid starting points XM and YN is based on the number of pixels existing in the macroblock.

For example, it is possible to move the X-axis and Y-axis grid starting points XM (YN) within the size of the X-axis and Y-axis of the macro block at intervals of unit pixels.

However, since the information on the color signal in the video signal is 1/2 of the information on the luminance signal, the movement interval K (X) of the X-axis and Y-axis grid start point XM (YN) L are preferably arranged at intervals of two unit pixels.

Also, in the above-described embodiment, it is explained that there is only one optimum grid start point in which the image of the object on which the VOP is formed exists in the smallest number of macro blocks. However, one or more optimal grid start points may occur.

Therefore, in the present invention, when a plurality of optimal grid start points occur, the macro block is subdivided into sub-blocks having a size of M / 2 x N / 2, and grid start points XM) YN by a predetermined interval (K) (L) and finds an information amount reduction position where the image of the object exists in the smallest number of macro blocks.

That is, when it is assumed that the size of a macroblock is 16 × 16 pixels, the sub-block is divided into sub-blocks having 8 × 8 pixels, and within the number of sub-blocks subdivided into 8 × 8 pixels, The grid start point XM is shifted by a predetermined distance K to find the X and Y axis grid start points having the smallest number of macroblocks of the object image and outputs it to the information amount decrease position.

A plurality of information amount reduction positions for counting the least number of macroblocks can be generated even if the granular macro block is subdivided into sub-blocks.

Therefore, in the present invention, any one of the X-axis and Y-side grid starting points (XOM, YON) optimal for the X-axis and Y-direction is selected even if subdivided into subblocks.

At this time, one information amount reduction position to be selected is a position where the amount of information decreases as the value of the motion vector decreases as the initial X-axis and Y-grid start points (XM = 0, YN = 0) Since the incidence of prediction error is low, one X-axis and Y-axis grid start point closest to the straight line distance is determined based on the initial grid start point (XM = 0, YN = 0).

As described above, according to the present invention, the shape information is encoded by adjusting the grid start point so that the image of the object is located in the minimum macro block, and the start point of the grid is adjusted so that the shape information encoding signal is located in the smallest macro block, And the object inside information are encoded, the coding efficiency is improved and the amount of information to be transmitted and stored is reduced.

No content

FIG. 1 is a block diagram showing the configuration of a VM encoder that is primarily determined by an organization under international standards.

FIG. 2 is a diagram showing a VOP having shape information divided into macroblocks according to a conventional method; FIG.

FIG. 3 is a block diagram showing the configuration of a VM decoder that is primarily determined by an organization under international standards.

FIG. 4 is a block diagram showing the configuration of a VOP coding unit which is primarily determined by an organization under the international standard.

FIG. 5 is a block diagram showing an embodiment in which a shape adaptive block division unit according to the encoding apparatus of the present invention is constituted in a VOP coding unit which is primarily determined by an international standard organization.

FIG. 6 is a circuit diagram showing an embodiment of the shape adaptive block dividing unit of FIG. 5; FIG.

7 is a circuit diagram showing an embodiment of the address generator of FIG. 6;

FIG. 8 illustrates an example of a process of outputting a macro block of an image stored in a memory of FIG. 6 according to an X-axis and a Y-axis address output from an address generator.

FIG. 9 is a circuit diagram showing the configuration of the block number counter in FIG. 6; FIG.

10 is a signal flow diagram showing a coding method of the present invention;

FIG. 11 is a diagram illustrating a state in which a macroblock of a VOP is moved to an information amount reduction position according to the encoding method of the present invention.

FIG. 12 is a signal flow diagram illustrating an embodiment of finding an information amount reduction position of a macroblock according to an encoding method of the present invention.

FIG. 13 is a signal flow diagram illustrating another embodiment for finding a position of a reduced amount of information of a macroblock according to the encoding method of the present invention. FIG.

FIG. 14 is a signal flow diagram illustrating yet another embodiment for finding a position of a reduced amount of information in a macroblock according to the coding method of the present invention. FIG.

15 (a) and 15 (b) show moving the grid start point of the macroblock zigzag according to another embodiment of FIG. 14;

FIG. 16 is a signal flow diagram illustrating yet another embodiment for finding a position of information reduction in a macroblock according to the encoding method of the present invention. FIG.

FIG. 17 is a signal flow diagram illustrating yet another embodiment for finding a position of a reduced amount of information in a macroblock according to the encoding method of the present invention. FIG.

FIG. 18 (B) shows a state in which a macroblock is moved to an information amount reduction position according to another embodiment of FIG. 16 and FIG. 17; FIG.

Explanation of reference numerals for the main sections of the drawings

11: VOP forming part, 12A, 12B, 12C, ... : VOP encoding unit, 13,38: multiplexer, 21,38: demultiplexer, 22A, 22B, 22C, ... A VOP decoder 23, a synthesizer 31, a motion estimator 32, a motion compensator 33, an adder 34, an object internal encoder 36, a previous VOP detector 37, The present invention relates to a method and apparatus for generating an address of a macroblock in a macroblock and a macroblock size selector for dividing the macroblock into blocks, 522: Y-axis size determination unit, 523: Block address generator, 541: Macro block counter, 542: Judgment unit, 543:

No content

No content

Claims (24)

A division step of dividing a VOP formed according to an image of an object into macroblocks having MxN pixels; a first search step of searching for a position of a reduced amount of information of a target object by moving a grid start point of a macroblock partitioned by the division step; A shape information coding step of coding shape information of an object image existing in a macroblock at an information amount reduction position found in the first search step; A motion estimation / encoding step of estimating and compensating motion in units of macroblocks at the information amount reduction position found in the second search step and encoding the object inside information in units of subblocks; Coded shape information and the motion estimation / And outputting the motion information and the coded object inside information. 2. The method according to claim 1, wherein the first and second search steps comprise: a moving step of moving the entire macroblocks along the X axis and / or the Y axis; And an output step of finding and outputting a position at which the amount of information is reduced in the determining step. 3. The method of claim 2, further comprising: a first movement counting step of counting the number of macroblocks in which an image of an object exists while moving the entire macroblocks along the X axis; A first setting step of setting an X-axis position of the macroblock in the X-axis position set in the first setting step, a macro block reconstruction step of setting a macroblock in the X-axis position set in the first setting step, A second setting step of setting a Y-axis position at which the smallest number of macroblocks is counted in the second movement counting step; and a second setting step of setting a Y- And a determination step of determining an X-axis and a Y-axis position as information amount reduction positions. 3. The method according to claim 2, further comprising: a first movement counting step of counting the number of macroblocks in which an image of an object is present while moving the entire macroblocks along the Y axis; A first setting step of setting a Y-axis position of the macroblock in the Y-axis position, a first setting step of setting a Y-axis position in the first setting step, A second moving step of moving the X-axis position where the least number of macroblocks are counted in the second movement counting step; and a second setting step of setting the X- And determining a Y-axis and an X-axis position as information amount reduction positions. 3. The method according to claim 2, further comprising: a first movement counting step of counting the number of macroblocks in which an image of an object exists while moving the entire macroblocks along the X axis; A second movement counting step of repeating the counting operation after the Y axis movement is completed in the Y axis and the Y axis movement in the first and second movement counting steps; And determining a Y-axis position as an information amount reduction position. 3. The method according to claim 2, further comprising: a first movement counting step of counting the number of macroblocks in which an image of an object exists while moving the entire macroblocks along the Y axis; And a second movement counting step of repeating the counting step after the X axis movement is completed on the X axis and the X axis movement in the first and second movement counting steps, And determining a Y-axis position as an information amount reduction position. 3. The method according to claim 2, further comprising: a moving counting step of counting macroblocks in which an image of an object is present while moving the entire macroblocks zigzag; and a step of determining a position at which the smallest number of macroblocks is counted in the counting step, Wherein the determination step comprises the steps of: The method according to claim 1, wherein the first = 1 and the second search step comprises: a moving step of moving the entire macroblock in the X-axis or the Y-axis; Or Y axis macroblock; and an output step of searching for and outputting an information amount reduction position in the determination step. 9. The method of claim 8, further comprising: a first counting step of counting macroblocks in which an image of an object exists while moving the entire macroblocks along the Y axis; And a second setting step of moving a macroblock of each row of the X axis along the X axis at the Y axis position set in the first setting step, A second setting step of setting a position where the smallest number of macroblocks is counted in the second counting step to a position of a row of each of the X axes; And determining a Y-axis position and a row position of each X-axis set in the setting step as an information amount reduction position. 9. The method according to claim 8, further comprising: a first counting step of counting macroblocks in which an image of an object exists while moving the entire macroblocks along the X axis; And a macro block of each column of the Y axis on which an image of the object is positioned while moving a macroblock of each column of the Y axis on the Y axis at the X axis position set in the first setting step, A second counting step of setting a position where the least number of macroblocks are counted in the second counting step to a position of each column of the Y axis; And determining a position of each of the set X-axis position and Y-axis position as an information amount reduction position. 9. The method according to claim 8, wherein the number of macroblocks in which an image of the object exists is counted while moving the macroblocks in each row on the X axis along the X axis, and the positions where the smallest number of macroblocks are counted are designated as rows Axis positioning step of moving all the macroblocks on the Y axis and repeating the X axis positioning step when the counting is completed in the X axis positioning step; A Y-axis positioning step of determining a Y-axis position where a value obtained by summing and summing the count values at the positions of the respective rows of the X-axis determined in the X-axis positioning step when the Y- The Y-axis position determined in the Y-axis positioning step and the position of each row of the X-axis obtained by counting the smallest macroblock in the determined Y-axis position are determined as the information amount reduction positions Wherein the determination step comprises the steps of: The method according to claim 8, wherein the number of macroblocks in which the image of the object exists is counted while moving the macroblocks in each column of the Y axis along the Y axis, and the position where the least number of macroblocks are counted is determined as the position of each column of the Y axis A step of moving the entire macroblock in the X axis and repeating the Y axis positioning step when the counting is completed in the Y axis position setting step; An X-axis position setting step of setting an X-axis position in which a sum obtained by summing and summing the count values at the positions of the respective Y-axis columns set in the Y-axis position setting step when the axis movement is completed, The X axis position set in the axis position setting step and the position of each of the Y axes where the smallest macroblock is counted in the set X side position is determined as the information amount reduction position Wherein the determination step comprises the steps of: The method of any one of claims 1 to 12, wherein the information amount reduction position is a position where the amount of information is minimum. The method of any one of claims 1 to 12, wherein the information amount reduction position is a position where a number of macroblocks in which an image of an object exists is a minimum position. The method of any one of claims 1 to 12, wherein the moving range of the macroblock is within the X-axis and Y-axis sizes of one macroblock. The method according to any one of claims 1 to 12, wherein the moving range of the macroblock is within the size of the X and Y axes of one macroblock, and moves to a unit pixel interval. The method according to any one of claims 1 to 12, wherein the moving range of the macroblock is within the size of the X axis and the Y axis of one macroblock, and moves at intervals of two unit pixels Way. The method as claimed in any one of claims 3 to 7 and 9 to 12, wherein when a plurality of X-axis and Y-axis positions counting the smallest number of macroblocks exist, Wherein the number of macroblocks in which an image of an object exists is counted and the position where the smallest number of macroblocks is counted is determined as an information amount reduction position. The method as claimed in claim 18, wherein, when a plurality of positions where the smallest number of macroblocks are counted after dividing the size of the X-axis and the Y-axis of the macroblock by 1/2, respectively, And the position of the object image is determined as the information amount reduction position. A first shape adaptive block dividing unit 40 for moving a macroblock of a VOP with respect to an object image formed by the VOP forming unit 11 to find an information amount reduction position, A shape coding unit 37 for coding the shape information of an object existing in a macroblock of the first shape adaptive block generating unit 37, (31) for estimating a motion of an object using output signals of the VOP forming unit (11) and the second shape adaptive block dividing unit (41); a motion estimating unit A motion compensator 32 for compensating for the motion of the object based on the motion information and the output signal of the second shape adaptive block dividing unit 41 and a motion compensator 32 for compensating for the output signal of the motion compensator 32 and the VOP forming unit 11. [ The difference value of the output signal of An object internal encoding unit 34 for encoding internal information of an object in accordance with an output signal of the adder 33 and an output signal of the second shape adaptive block dividing unit 41, An adder 35 for adding the output signals of the compensating unit 32 and the object inner coding unit 34 and an adder 35 for detecting the V0P of the previous picture with the output signal of the adder 35, A previous V0P detection unit 36 for use in motion estimation and motion compensation in the motion compensation unit 32 and motion estimation means 36 for estimating motion information estimated by the motion estimation unit 31, And a multiplexer (38) for outputting the internal information of the object image and the shape information encoded by the shape encoding unit (37). The apparatus as claimed in claim 20, wherein the first and second shape adaptive block dividing units (40, 41) include: address generation control means for outputting a start position for generating an address by moving a predetermined distance within a size of a macro block, An address generating means for generating an address for outputting an image of an object as a macro block in accordance with an address start position output by the control means; and a control means for storing an image of the object having the input shape information, A block number counting means for counting the number of macroblocks in which the shape information of the object exists among the signals output from the memory means; The grid start of the X and Y axes with the smallest number of macroblocks counted Of the object to the image encoding apparatus, characterized by at least consists of a macroblock grid selecting means for selecting values. 22. The image processing apparatus according to claim 21, wherein the address generating means comprises X-axis size determining means and Y-axis size determining means for respectively determining the X-axis size of the macro block and the Y-axis size of the macro block in accordance with the size information of the input object image, Axis size and Y-axis size of the macroblock determined by the X-axis size determination means and the Y-axis size determination means from the address start position output from the address generation control means, And an area address generating means for sequentially generating an address in accordance with the address information. The apparatus of claim 22, wherein the X-axis and Y-axis sizes of the macroblock are determined by one size determining means when the X-axis and Y-axis sizes of the macroblock are the same. The apparatus according to claim 21, wherein the block number counting means comprises: area counting means for counting clock signals to divide macroblocks; and macroblocks of images output from the memory means are classified according to output signals of the area counting means, And a block number adding means for counting the judgment signal of the shape information existence judging means and outputting the number of macroblocks in which an image of the object exists, Of the object image. ※ Note: It is disclosed by the contents of the first application.